Simpson’s paradox might have omicron appear less deadly even if it is more deadly
A reader suggested that Tomas chose extreme numbers to "make the Paradox" work. So I repeated the calculation with less extreme numbers, and the Simpson Paradox is still there
It matters! Because if you were already immune, your chances of dying just went up.

(and so did if you weren’t)
This further exacerbates Simpson’s paradox

• • •

Missing some Tweet in this thread? You can try to force a refresh
 

Keep Current with Luca Dellanna

Luca Dellanna Profile picture

Stay in touch and get notified when new unrolls are available from this author!

Read all threads

This Thread may be Removed Anytime!

PDF

Twitter may remove this content at anytime! Save it as PDF for later use!

Try unrolling a thread yourself!

how to unroll video
  1. Follow @ThreadReaderApp to mention us!

  2. From a Twitter thread mention us with a keyword "unroll"
@threadreaderapp unroll

Practice here first or read more on our help page!

More from @DellAnnaLuca

13 Dec
OMICRON EXPECTED DEATHS

I'm reading news such as "no one died of omicron in Europe yet, so it's milder."

However, omicron is recent, so is the lack of deaths an artifact of the simple fact that it takes time to die?

Let's figure it out by running some numbers.
Most of the current omicron cases got infected recently, so even those who would die eventually are not died yet.

The median delay between COVID infection and death is 11 days (computed pre-vaccine). Let's assume is correct.
Today is the 13th, so we want to know how many cases there were on the 2nd of December. According to ECDC, there were 79 cases.

That said, there might have been more cases than detected; let's say 4x. Total infected: 500.
Read 9 tweets
4 Dec
100 out of 120 participants to a Christmas dinner infected with Omicron. They were all vaccinated and tested negative before the dinner (according to the company). One attendee was back from SA 🇿🇦

A mild super-spreading event.

Source: expressen.se/nyheter/sa-ble… via @gianlucac1
"The person who is believed to have caused the spread of infection is an employee who returned from South Africa a few days earlier."
Read 5 tweets
27 Nov
💯 even if true that variants get trade off lethality for transmissibility over time, a hypothetical variant that is 20% less lethal but 20% more contagious kills more people than the original virus (because cases compound → larger exposure → more total deaths)
(I didn’t check the estimate Giullaume quoted; here, I was just commenting on his general point.)
That said, I question the hypothesis that variants always trade off lethality for transmissibility. While I understand that more transmissible variants get selected, I don’t see why a variant couldn’t be both more transmissible and more lethal.
Read 7 tweets
25 Nov
Highest-ever number of cases in Europe. Highest-ever virus attack surface.

If you’re not vaccinated, know that the benefit-costs ratio increased since January.

And if you’re vaccinated, your absolute risk didn’t decrease as much as the nominal 80-90% vaccine protection.
Yes. Waves are partially due to seasonality, partially due to immunity fading / variants, and partially due to the virus filling a network fast📈, reaching herd immunity there📉, then finding access to another network📈
In the first tweet, by “the benefit/cost ratio increased”, I meant of vaccination. In other words, it’s more risky not to vaccinate now that it was a year ago, and it’s less risky to vaccinate now than it was in January.

(HT @maintcraft for pointing out the potential ambiguity)
Read 4 tweets
22 Nov
This definitely matters (though I don’t know how much) and is usually missed by most models which – usual mistake – consider the population homogeneous.
What do I mean, “how much”?

If you asked if it matters, I’d say yes. I’d guess it explains a high percentage of wave behavior.

But if you asked me to list all causes, assign a percentage to each, and then normalize so that the total is 100%, I don’t think it’d be a very high %.
I hear that some think that an heterogeneous population has a lower herd immunity threshold, because the few superconnectors get infected first and stop connecting the groups.

But the superconnectors aren’t the only link between groups!

(continues below)
Read 4 tweets
22 Nov
I don’t know who needs to hear that, but:

- daily cases in the Netherlands are 2x the max last year

→ the 25% unvaccinated in the Netherlands are ~double as likely to die as 1y ago; prob more due to Delta

→ the 75% vaccinated have < half the risk reduction ~expected 1y ago
And probably reality is worse due to not having reach the top yet (probably)
For completeness: OTOH, the 75% vaccinated means that vaccinated cases contribute less to spread; OTOH, the asymptomatic vaccinated is also less likely to test, so in the numbers above I imagine that the two effects above cancel each other.

More in general:
Read 5 tweets

Did Thread Reader help you today?

Support us! We are indie developers!


This site is made by just two indie developers on a laptop doing marketing, support and development! Read more about the story.

Become a Premium Member ($3/month or $30/year) and get exclusive features!

Become Premium

Too expensive? Make a small donation by buying us coffee ($5) or help with server cost ($10)

Donate via Paypal

Or Donate anonymously using crypto!

Ethereum

0xfe58350B80634f60Fa6Dc149a72b4DFbc17D341E copy

Bitcoin

3ATGMxNzCUFzxpMCHL5sWSt4DVtS8UqXpi copy

Thank you for your support!

Follow Us on Twitter!

:(